OK, you mention that you are following an established protocol, so these suggestions should already be part of that protocol but I'll mention them anyway.

You should add protease inhibitors (e.g. Roche EDTA-free tablets) to your lysis buffer before you lyse the cells.

I believe your protein has a pI of approximately 6, so all your buffers should be in the pH range 7.6 to 8.0. Use buffers at a concentration of 100 mM rather than 20-50 mM, as the pH is quite important and has to be stable during most affinity chromatography steps. You can reduce the concentration when you dialyse your sample after the affinity step.

Imidazole will affect the pH of your buffer, so you need to pH the lysis, binding and elution buffers AFTER adding all the components to them. I know somebody who used to make a few 1M stock solutions of TRIS with different pH's and then, when he needed to make a buffer he just added the necessary NaCl, imidazole and H2O to the stock solution and assumed the resulting buffer would still be at the same pH as the stock. It obviously wasn't. Imidazole is a buffer by itself and if you have 300-500 mM imidazole in your buffer and only 20-50 mM TRIS, then the imidazole will have a strong effect on the pH. Also, the pH of TRIS changes with temperature, so make sure you check it if your purification temperature is different from the one you used to make the buffer.

Your binding buffer should have 10-30 mM imidazole in it to reduce non-specific binding of contaminating proteins to your Ni-NTA column or beads. 500 mM NaCl also helps with reducing non-specific binding and keeping your protein in solution, as will 5-10% glycerol. You can dialyse these out afterwards.

Add a reducing agent to your buffer to remove contaminants that might bind to you protein and to stop your protein oligomerising if it is capable of forming disulphide bonds via cysteines (monomers tend to purifier better). 5mM betamercaptoethanol is good for Ni-NTA, stronger reducing agents might reduce the nickel ions from the affinity matrix and ruin you purification,

If you use batch elution rather than a gradient, you should introduce at least a couple of wash steps to remove as many contaminants as possible before eluting with full strength elution buffer (e.g. one wash with 50 mM imidazole buffer and one with 70 mM imidazole). The cleaner your eluted protein, the less likely it will be to precipitate.

Some proteins don't like high imidazole concentrations and will stay in solution more readily during dialysis if your elution buffer was in the 200 to 300 mM imidazole range rather than >500 mM before dialysis. Other proteins need high salt concentrations to stay in solution and crash out if you reduce the NaCl concentration too much during dialysis.

Some nickel might leach from your beads or column and His-tagged proteins can aggregate and precipitate because their tags bind to the free-floating nickel ions. Adding 1 mM EDTA to your dialysis buffer will remove the nickel and you can remove this via a second dialysis step if you need to. Don't use EDTA if your protein is bound to a metal co-factor, as it will chelate these metal ions as well as the nickel and that might make your protein structurally unstable and precipitate.

Finally, one of my proteins used to precipitate if I dialysed it while it was at a concentration >0.4 mg/ml after Ni-NTA. Reducing its concentration with dialysis buffer right after elution and before putting it into dialysis helped a lot in this case.

Dear Nuria,

First few thoughts are:

- your protein is concentrated during to long dialysis if you have glycerol outside the bag

- you can have crystallization not aggregation, you can just look at it under the microscope. Spun-down crystals would be the purest version of your protein you can imagine.

- you have your sample contaminated by the aggregates promote further aggregation. High G long spun down should help in this case

Keep fingers cross - Jan

OK, you mention that you are following an established protocol, so these suggestions should already be part of that protocol but I'll mention them anyway.

You should add protease inhibitors (e.g. Roche EDTA-free tablets) to your lysis buffer before you lyse the cells.

I believe your protein has a pI of approximately 6, so all your buffers should be in the pH range 7.6 to 8.0. Use buffers at a concentration of 100 mM rather than 20-50 mM, as the pH is quite important and has to be stable during most affinity chromatography steps. You can reduce the concentration when you dialyse your sample after the affinity step.

Imidazole will affect the pH of your buffer, so you need to pH the lysis, binding and elution buffers AFTER adding all the components to them. I know somebody who used to make a few 1M stock solutions of TRIS with different pH's and then, when he needed to make a buffer he just added the necessary NaCl, imidazole and H2O to the stock solution and assumed the resulting buffer would still be at the same pH as the stock. It obviously wasn't. Imidazole is a buffer by itself and if you have 300-500 mM imidazole in your buffer and only 20-50 mM TRIS, then the imidazole will have a strong effect on the pH. Also, the pH of TRIS changes with temperature, so make sure you check it if your purification temperature is different from the one you used to make the buffer.

Your binding buffer should have 10-30 mM imidazole in it to reduce non-specific binding of contaminating proteins to your Ni-NTA column or beads. 500 mM NaCl also helps with reducing non-specific binding and keeping your protein in solution, as will 5-10% glycerol. You can dialyse these out afterwards.

Add a reducing agent to your buffer to remove contaminants that might bind to you protein and to stop your protein oligomerising if it is capable of forming disulphide bonds via cysteines (monomers tend to purifier better). 5mM betamercaptoethanol is good for Ni-NTA, stronger reducing agents might reduce the nickel ions from the affinity matrix and ruin you purification,

If you use batch elution rather than a gradient, you should introduce at least a couple of wash steps to remove as many contaminants as possible before eluting with full strength elution buffer (e.g. one wash with 50 mM imidazole buffer and one with 70 mM imidazole). The cleaner your eluted protein, the less likely it will be to precipitate.

Some proteins don't like high imidazole concentrations and will stay in solution more readily during dialysis if your elution buffer was in the 200 to 300 mM imidazole range rather than >500 mM before dialysis. Other proteins need high salt concentrations to stay in solution and crash out if you reduce the NaCl concentration too much during dialysis.

Some nickel might leach from your beads or column and His-tagged proteins can aggregate and precipitate because their tags bind to the free-floating nickel ions. Adding 1 mM EDTA to your dialysis buffer will remove the nickel and you can remove this via a second dialysis step if you need to. Don't use EDTA if your protein is bound to a metal co-factor, as it will chelate these metal ions as well as the nickel and that might make your protein structurally unstable and precipitate.

Finally, one of my proteins used to precipitate if I dialysed it while it was at a concentration >0.4 mg/ml after Ni-NTA. Reducing its concentration with dialysis buffer right after elution and before putting it into dialysis helped a lot in this case.

In addition to all good comments of Antonio, I do think that your protein does not like long time of dialysis (most likely for one of the reasons advanced by Antonio). I had a lot of success with MSPs proteins with desalting columns (like Hi Trap). Its fast and efficient and much better for the protein. I always do one step affinity and then desalting and the MSP protein is ready to use for doing nice nanodiscs. 

Good luck and cheers

Thank you both, we have solved the problem adding a small amount of cholate to the sample, it seems that it increases protein solubility at least during sample concentration and we there is no problem for us using it in that conditions.